摘要
构建了一个图像识别及检测的DSP+FPGA硬件平台,采用机器视觉技术,对烟包外包装各种信息进行准确而快速的识别。设计了一个烟包封签缺陷检测算法,选取图像强度分量进行灰度处理,选择合适的二值化和滤噪方案对图像进行二值化及滤噪处理,最终在二值化图像中进行封签的偏移、叠角和缺失的检测。为烟包的高速自动化包装检测提供了一种高效可行的实施方案。
A hardware platform with DSP＋FPGA is constructed to identify and supervise images. Machine vision is used to identify the information of cigarette packages. An algorithm of faults supervision of cigarette tag paper is designed. The intensity of images should be firstly processed by shade, then denoise and binarization processes are carried out. The faults of package tag paper such as offset, overlap angle, etc. are identified. This method is efficient in automatic package supervision with high speed.
出处
《工程图学学报》
CSCD
北大核心
2009年第3期131-135,共5页
Journal of Engineering Graphics
基金
河南省重点科技公关资助项目(072102210013)
关键词
计算机应用
图像处理
封签检测
机器视觉
烟包
computer application
image processing
tag paper supervision
machine vision
cigarette
